Spatial Stratification and Similarity Surfaces for Soil Survey

Applications of the rassta package with SSURGO data

Unsupervised Hillslope Classification

The issue:

Lack of high-resolution spatial representations of mental soil-landscape models

The opportunity:

Unsupervised hillslope classification through robust unsupervised learning with rassta

Unsupervised Hillslope Classification

Terrain attributes selected through pedologic reasoning

Terrain attributes selected through pedologic reasoning

Unsupervised Hillslope Classification

Self-organizing map with rassta::somgap()Self-organizing map with rassta::somgap()

Self-organizing map with rassta::somgap()

Unsupervised Hillslope Classification

Selection of k with rassta::somgap()Selection of k with rassta::somgap()

Selection of k with rassta::somgap()

Unsupervised Hillslope Classification

Agglomerative Hierarchical Clustering of the Self-organizing map

Agglomerative Hierarchical Clustering of the Self-organizing map

Unsupervised Hillslope Classification

Spatial representation of the self-organizing map and its clusters (i.e., hillslope positions) with rassta::sompam()Spatial representation of the self-organizing map and its clusters (i.e., hillslope positions) with rassta::sompam()

Spatial representation of the self-organizing map and its clusters (i.e., hillslope positions) with rassta::sompam()

Thanks